Fig. 2
Restricted randomization of the trial. This figure shows how we carried out restricted randomization for the study. We stratified quartiers depending on the child diarrheal prevalence, and within each stratum we grouped the quartiers into two (shown graded and ungraded in the figure) in order to ensure that quartiers in the same villages are allocated to the same arm and also to allocate the same number of quartiers to the treatment and control arms. Quartiers in the green boxes were selected as representative for each group in each stratum, and the leaders from the quartiers participated in the randomization activity, selecting an envelope containing a paper marked O or X without knowing which it would be before they opened it. For instance, if the leader of Quartier 1 selected an envelope with O, all the quartiers from Q1 through Q5 were to be allocated to the treatment group and Q6 through Q10 were remaining for the control group. Using this method, there were 8 different possible allocation scenarios for the trial. (Village and quartier name) V1: Bangabanga, V2: Bwalenge, V3: KalangandaMukeni, V4: Mayanda, V5: Ingundu, V6: ImpiniNnsi, V7: Punkulu, V8: IntswemLabwi, Q1: Bangabanga1, Q2: Bangabanga2, Q3: MbuluEbeth, Q4: Mbulesal, Q5: Ibansi, Q6: Center, Q7: Lakam, Q8: Ndjili, Q9: Nganda, Q10: Orondi, Q 11: Camp Center de Sante, Q12: Camp Musanga, Q13: Ekong, Q14: ImbilEngwow, Q15: Tshikapa, Q16: BwalaNgundu, Q17: Eban, Q18: Sante Labwi2